Identification of Animal Species from Their Sounds

摘要

Identification of animal species based on their sounds has already proven to be useful in biodiversity assessment. In this paper we explore the use of a combined Teager- cepstral-TESPAR (Time Encoded Signal Processing And Recognition) analysis to discriminate between different animal species. Our experiments using this approach together with classification techniques shows that TESPAR S-matrices of Teager cepstral coefficients along with some additional features can be successfully used to discriminate between different animal species, even in the conditions of small training sets.